library(tidyverse)
library(ggfortify)
library(GGally)
library(skimr)
library(modelr)

wine_quality_red <- read_csv("data/wine_quality_red.csv")

Rows: 1599 Columns: 14── Column specification ────────────────────────────────────────────────────────────────────────────────
Delimiter: ","
chr  (1): region
dbl (13): wine_id, fixed_acidity, volatile_acidity, citric_acid, residual_sugar, chlorides, free_sul...
ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

wine_quality_white <- read_csv("data/wine_quality_white.csv")

Rows: 4898 Columns: 14── Column specification ────────────────────────────────────────────────────────────────────────────────
Delimiter: ","
chr  (1): region
dbl (13): wine_id, fixed_acidity, volatile_acidity, citric_acid, residual_sugar, chlorides, free_sul...
ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

head(wine_quality_red)

skim(wine_quality_red)

── Data Summary ────────────────────────
                           Values          
Name                       wine_quality_red
Number of rows             1599            
Number of columns          14              
_______________________                    
Column type frequency:                     
  character                1               
  numeric                  13              
________________________                   
Group variables            None            

wine_quality_red %>% 
  distinct(region)

head(wine_quality_white)

skim(wine_quality_white)

── Data Summary ────────────────────────
                           Values            
Name                       wine_quality_white
Number of rows             4898              
Number of columns          14                
_______________________                      
Column type frequency:                       
  character                1                 
  numeric                  13                
________________________                     
Group variables            None              

wine_quality_white %>% 
  distinct(region)

wine_quality_red %>% 
  ggplot(aes(x = quality)) +
  geom_histogram()

wine_quality_white %>% 
  ggplot(aes(x = quality)) +
  geom_histogram()

ggpairs(wine_quality_red %>% 
          select(-wine_id,
                 -region))

quality ~ volatile_acidity, sulphates, alcohol

ggpairs(wine_quality_white %>% 
          select(-wine_id,
                 -region),
        progress = FALSE)

quality ~ alcohol, chlorides, density

model_r1a <- lm(quality ~ volatile_acidity,
                data = wine_quality_red)

model_r1b <- lm(quality ~ sulphates,
                data = wine_quality_red)

model_r1c <- lm(quality ~ alcohol,
                data = wine_quality_red)

model_w1a <- lm(quality ~ alcohol,
                data = wine_quality_white)

model_w1b <- lm(quality ~ chlorides,
                data = wine_quality_white)

model_w1c <- lm(quality ~ density,
                data = wine_quality_white)

summary(model_r1a)

Call:
lm(formula = quality ~ volatile_acidity, data = wine_quality_red)

Residuals:
     Min       1Q   Median       3Q      Max 
-2.74440 -0.54813 -0.02734  0.53280  2.77353 

Coefficients:
                 Estimate Std. Error t value Pr(>|t|)    
(Intercept)       6.57413    0.06155  106.81   <2e-16 ***
volatile_acidity -1.72666    0.11044  -15.63   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.7905 on 1597 degrees of freedom
Multiple R-squared:  0.1327,    Adjusted R-squared:  0.1322 
F-statistic: 244.4 on 1 and 1597 DF,  p-value: < 2.2e-16

summary(model_r1b)

Call:
lm(formula = quality ~ sulphates, data = wine_quality_red)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.5954 -0.6056 -0.0354  0.5438  2.7696 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  4.90612    0.08286  59.209   <2e-16 ***
sulphates    1.14965    0.12192   9.429   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.8262 on 1597 degrees of freedom
Multiple R-squared:  0.05274,   Adjusted R-squared:  0.05214 
F-statistic: 88.91 on 1 and 1597 DF,  p-value: < 2.2e-16

summary(model_r1c)

Call:
lm(formula = quality ~ alcohol, data = wine_quality_red)

Residuals:
     Min       1Q   Median       3Q      Max 
-2.95517 -0.50166 -0.04501  0.49519  2.70011 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  1.89570    0.18603   10.19   <2e-16 ***
alcohol      0.36142    0.01776   20.36   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.7564 on 1597 degrees of freedom
Multiple R-squared:  0.206, Adjusted R-squared:  0.2055 
F-statistic: 414.4 on 1 and 1597 DF,  p-value: < 2.2e-16

alcohol is best for red wrt quality

summary(model_w1a)

Call:
lm(formula = quality ~ alcohol, data = wine_quality_white)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.5247 -0.5623 -0.0207  0.5432  3.2832 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept) 2.523661   0.104913   24.05   <2e-16 ***
alcohol     0.321456   0.009911   32.44   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.8535 on 4896 degrees of freedom
Multiple R-squared:  0.1769,    Adjusted R-squared:  0.1767 
F-statistic:  1052 on 1 and 4896 DF,  p-value: < 2.2e-16

summary(model_w1b)

Call:
lm(formula = quality ~ chlorides, data = wine_quality_white)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.3853 -0.6165  0.0017  0.5394  3.4159 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  6.30427    0.03055  206.33   <2e-16 ***
chlorides   -8.75508    0.60243  -14.53   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.921 on 4896 degrees of freedom
Multiple R-squared:  0.04135,   Adjusted R-squared:  0.04116 
F-statistic: 211.2 on 1 and 4896 DF,  p-value: < 2.2e-16

summary(model_w1c)

Call:
lm(formula = quality ~ density, data = wine_quality_white)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.3995 -0.6006  0.0039  0.5651  3.9017 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)   98.452      4.267   23.07   <2e-16 ***
density      -93.104      4.293  -21.69   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.8985 on 4896 degrees of freedom
Multiple R-squared:  0.08764,   Adjusted R-squared:  0.08746 
F-statistic: 470.3 on 1 and 4896 DF,  p-value: < 2.2e-16

alcohol best for white wrt quality

red_resid <- wine_quality_red %>% 
  add_residuals(model = model_r1c) %>% 
  select(-region, -quality, -alcohol)

white_resid <- wine_quality_white %>% 
  add_residuals(model = model_w1a) %>% 
  select(-region, -quality, -alcohol)

ggpairs(red_resid,
        progress = FALSE)

volatile_acidity, sulphates

ggpairs(white_resid,
        progress = FALSE)

volatile_acidity, free_sulfur_dioxide

model_r2a <- lm(formula = quality ~ alcohol + volatile_acidity,
                data = wine_quality_red)

model_r2b <- lm(formula = quality ~ alcohol + sulphates,
                data = wine_quality_red)

model_w2a <- lm(formula = quality ~ alcohol + volatile_acidity,
                data = wine_quality_white)

model_w2b <- lm(formula = quality ~ alcohol + free_sulfur_dioxide,
                data = wine_quality_white)

summary(model_r2a)

Call:
lm(formula = quality ~ alcohol + volatile_acidity, data = wine_quality_red)

Residuals:
     Min       1Q   Median       3Q      Max 
-2.55192 -0.47030 -0.01778  0.47927  2.41435 

Coefficients:
                 Estimate Std. Error t value Pr(>|t|)    
(Intercept)       3.08357    0.19857   15.53   <2e-16 ***
alcohol           0.31565    0.01723   18.32   <2e-16 ***
volatile_acidity -1.34665    0.10254  -13.13   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.7188 on 1596 degrees of freedom
Multiple R-squared:  0.2834,    Adjusted R-squared:  0.2825 
F-statistic: 315.7 on 2 and 1596 DF,  p-value: < 2.2e-16

summary(model_r2b)

Call:
lm(formula = quality ~ alcohol + sulphates, data = wine_quality_red)

Residuals:
    Min      1Q  Median      3Q     Max 
-2.9659 -0.4688 -0.0193  0.4856  2.3045 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  1.42025    0.19007   7.472 1.29e-13 ***
alcohol      0.34735    0.01744  19.919  < 2e-16 ***
sulphates    0.94527    0.10963   8.623  < 2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.7396 on 1596 degrees of freedom
Multiple R-squared:  0.2413,    Adjusted R-squared:  0.2404 
F-statistic: 253.9 on 2 and 1596 DF,  p-value: < 2.2e-16

volatile_acidity for new red_ model_resid

anova(model_r1c, model_r2a)

Analysis of Variance Table

Model 1: quality ~ alcohol
Model 2: quality ~ alcohol + volatile_acidity
  Res.Df    RSS Df Sum of Sq      F    Pr(>F)    
1   1597 913.65                                  
2   1596 824.54  1    89.112 172.49 < 2.2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

red_resid <- wine_quality_red %>% 
  add_residuals(model = model_r2a, "resid") %>% 
  select(-wine_id, -quality, -alcohol, -volatile_acidity)

summary(model_w2a)

Call:
lm(formula = quality ~ alcohol + volatile_acidity, data = wine_quality_white)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.6165 -0.5479 -0.0125  0.5235  3.2676 

Coefficients:
                 Estimate Std. Error t value Pr(>|t|)    
(Intercept)       2.97427    0.10506   28.31   <2e-16 ***
alcohol           0.33282    0.00964   34.52   <2e-16 ***
volatile_acidity -2.04894    0.11770  -17.41   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.8283 on 4895 degrees of freedom
Multiple R-squared:  0.2249,    Adjusted R-squared:  0.2245 
F-statistic:   710 on 2 and 4895 DF,  p-value: < 2.2e-16

summary(model_w2b)

Call:
lm(formula = quality ~ alcohol + free_sulfur_dioxide, data = wine_quality_white)

Residuals:
    Min      1Q  Median      3Q     Max 
-4.2055 -0.5615 -0.0305  0.5487  3.3317 

Coefficients:
                     Estimate Std. Error t value Pr(>|t|)    
(Intercept)         2.0726612  0.1167127  17.759   <2e-16 ***
alcohol             0.3432123  0.0101611  33.777   <2e-16 ***
free_sulfur_dioxide 0.0062945  0.0007352   8.561   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.8472 on 4895 degrees of freedom
Multiple R-squared:  0.189, Adjusted R-squared:  0.1887 
F-statistic: 570.5 on 2 and 4895 DF,  p-value: < 2.2e-16

volatile_acidity

anova(model_w1a, model_w2a)

Analysis of Variance Table

Model 1: quality ~ alcohol
Model 2: quality ~ alcohol + volatile_acidity
  Res.Df    RSS Df Sum of Sq      F    Pr(>F)    
1   4896 3566.3                                  
2   4895 3358.4  1    207.91 303.04 < 2.2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

white_resid <- wine_quality_white %>% 
  add_residuals(model_w2a, "white_resid") %>% 
  select(-wine_id, -quality, -alcohol, -volatile_acidity)
  

ggpairs(red_resid,
        progress = FALSE)

will add sulphates and compare adjr^2

sulphates is significant

summary(model_r3)

Call:
lm(formula = quality ~ alcohol + volatile_acidity + sulphates, 
    data = wine_quality_red)

Residuals:
     Min       1Q   Median       3Q      Max 
-2.66345 -0.45393 -0.01326  0.48807  2.16111 

Coefficients:
                 Estimate Std. Error t value Pr(>|t|)    
(Intercept)       2.62881    0.21133  12.439  < 2e-16 ***
alcohol           0.31134    0.01707  18.241  < 2e-16 ***
volatile_acidity -1.19442    0.10476 -11.401  < 2e-16 ***
sulphates         0.63716    0.10886   5.853 5.84e-09 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.7114 on 1595 degrees of freedom
Multiple R-squared:  0.2985,    Adjusted R-squared:  0.2972 
F-statistic: 226.2 on 3 and 1595 DF,  p-value: < 2.2e-16

adjr^2 goes up so will add

red_resid <- wine_quality_red %>% 
  add_residuals(model_r3, "red_resid") %>% 
  select(-wine_id, -quality, -alcohol, -volatile_acidity, -sulphates)

will try residual sugar

summary(model_w3) 

Call:
lm(formula = quality ~ alcohol + volatile_acidity + residual_sugar, 
    data = wine_quality_white)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.6037 -0.5389 -0.0167  0.5150  3.1930 

Coefficients:
                  Estimate Std. Error t value Pr(>|t|)    
(Intercept)       2.304082   0.122414   18.82   <2e-16 ***
alcohol           0.383596   0.010722   35.77   <2e-16 ***
volatile_acidity -2.178114   0.117109  -18.60   <2e-16 ***
residual_sugar    0.026953   0.002601   10.36   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.8194 on 4894 degrees of freedom
Multiple R-squared:  0.2415,    Adjusted R-squared:  0.241 
F-statistic: 519.4 on 3 and 4894 DF,  p-value: < 2.2e-16

summary(model_w2a)

Call:
lm(formula = quality ~ alcohol + volatile_acidity, data = wine_quality_white)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.6165 -0.5479 -0.0125  0.5235  3.2676 

Coefficients:
                 Estimate Std. Error t value Pr(>|t|)    
(Intercept)       2.97427    0.10506   28.31   <2e-16 ***
alcohol           0.33282    0.00964   34.52   <2e-16 ***
volatile_acidity -2.04894    0.11770  -17.41   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.8283 on 4895 degrees of freedom
Multiple R-squared:  0.2249,    Adjusted R-squared:  0.2245 
F-statistic:   710 on 2 and 4895 DF,  p-value: < 2.2e-16

residual sugar passes

white_resid <- wine_quality_white %>% 
  add_residuals(model = model_w3, "white_resid") %>% 
  select(-wine_id, -quality, -alcohol, -volatile_acidity, -resiudal_sugar)

Error in `select()`:
! Can't subset columns that don't exist.
✖ Column `resiudal_sugar` doesn't exist.
Backtrace:
 1. ... %>% ...
 3. dplyr:::select.data.frame(., -wine_id, -quality, -alcohol, -volatile_acidity, -resiudal_sugar)